A unique sodium-hydrogen exchange isoform (NHE-4) of the inner medulla of the rat kidney is induced by hyperosmolarity.

Membrane sodium-hydrogen exchangers (NHEs), found in virtually all cell types, appear to have diverse and essential roles in regulating cellular pH and mediating vectorial transport by epithelial cells. However, the functional and physiological role of the recently cloned isoform NHE-4 remains unknown. Unlike other Na-H exchanger isoforms, NHE-4 transfected into NHE-deficient mutant fibroblasts demonstrated no amiloride-inhibitable sodium uptake, under basal or acid-loaded isoosmotic conditions. By immunoblot analysis, only the NHE-4 transfectants synthesized a 100-kDa protein, which cross-reacted to polyclonal antibody made to an NHE-4 fusion protein. However, when cells were subjected to acute hyperosmolar cell shrinkage conditions, amiloride-sensitive NHE activity was readily detected at 420 mosm, exhibiting maximal activity at 490 mosm. By in situ hybridization, NHE-4 expression in the rat kidney was found to be limited to the inner renal medullary collecting tubules, the region of highest tissue osmolarity fluctuations in the body. We conclude that NHE-4 is an unusual isoform of sodium-hydrogen exchangers that may play a specialized supplementary role in cell volume regulation.

In contrast to NHE-1, the other isoforms remain incompletely characterized and have undefined physiological roles.
Preliminary studies have suggested that NHE-2 and NHE-3 are selectively, but not exclusively, expressed by epithelial cells of tissues such as intestine and kidney (4-7). NHE-3 is localized on the apical membrane of the small intestine (9). Thus, these two isoforms are potential mediators of vectorial sodium transport, a major function of such cells.
Far less is known about NHE-4. By Northern blot analyses, it appears to be most highly expressed in stomach, intestine, kidney, brain, uterus, and skeletal muscle (4), but its functional characteristics, membrane localization, and physiological role(s) are unknown. In this report, we provide evidence that expression of NHE-4 in the kidney is restricted to the collecting tubules of the inner medulla, the region of highest tissue osmolarity of the body. We also demonstrate that transfected full-length NHE-4 is quiescent in exchanger-deficient fibroblasts when acid-loaded at isoosmolarity but is activated when acid-loaded under hyperosmolar conditions. We speculate that this isoform may play a specialized role in the kidney in rectifying cell volume in response to extreme fluctuations of hyperosmolar-stimulated cell shrinkage.
The cRNA riboprobe (pG7N4A) was constructed by ligating the unique BarnHI,,-EcoRV,, NHE-4 cDNA fragment into BamHIISmaIcut pGEM7Z (Promega). SP6 and T7 promoters were used to generate, respectively, sense and antisense cRNA probes. This construct generated a probe that included 428 bases of 5' non-coding region and 166 bases of coding region corresponding to amino acids 1-56 of NHE-4.
A fusion protein to glutathione S-transferase consisted of the fragment BstBYI,,,,,, ligated into the BamHI-cut vector pGEX-3X (Pharmacia Biotech Inc.). This generated a fusion encompassing amino acids 396625 and was verified by Sanger dideoxy sequencing.
Cell Cnlture-PS120 (NHE-deficient) and PS127 (human NHE-1transfected) fibroblast cells, Chinese hamster lung CC139 derivatives (10) were provided to M. Villereal by J. Pouyssegur. Cells were maintained at 37 "C under an atmosphere of 5% CO,, 95% air in Dulbecco's modified Eagle's medium (high glucose) supplemented with 10% fetal calf serum (Life Technologies, Inc.), 5 unitdm1 penicillin, and 5 pg/ml streptomycin. RNA and protein was prepared from pre-and post-confluent cells according to previously described methods. Transfections into the PS120 cell line were done with Lipofectin (Life Technologies, Inc.), according to manufacturer's directions. All NHE-4 transfectants were selected by resistance to the antibiotic Geneticin (G418) (Life Technologies, Inc.), a gene camed on the shuttle vector pCB6+. G418resistant cells were clonally selected, NHE-4 expression was verified by Northern and Western analysis, and four clones were chosen for further analysis. The clone used for expression studies here, PSCN4-4, is representative of these.
Preparation of RNA and Ribonuclease Protection Assay-Isolation and purification of RNA were as described previously (9). Ribonuclease protection was done as described in the Promega Biotech Technical Bulletin. For the antisense strand probe, pG7N4A was linearized by BamHI (Life Technologies, Inc.) and the 607-base cRNA probe was generated by T7 polymerase (Promega). SP6 polymerase transcribed the 652-base sense probe from EcoRI-linearized pG7N4A. [ C Y -~~P~C T P or -UTP and [a-"SlUTP, respectively, were used to label internally the cRNA for ribonuclease protection assays and in situ hybridizations.
In Situ Hybridizations-Male Sprague-Dawley rats were deeply anesthetized by an intraperitoneal injection of 7% chloral hydrate and were then transcardially perfused with a solution of 4% paraformaldehyde (initially a t a pH of 6.5, then changed to the same solution a t pH 9.5). Tissues were resected and placed in a solution of 15% sucrose and 4% paraformaldehyde overnight. Cryostat sections (12 pm) were mounted on gelatin-and poly-L-lysine-coated slides. The slides were treated with a solution of 0.001% proteinase K i n 0.1 M Tris-HC1, pH 8, and 0.05 M EDTA, pH 8, for 30 min a t 37 "C and then with 0.025% acetic anhydride for 10 min. They were rinsed in 2 x SSC and dehydrated in increasing concentrations of ethanol. The tissue sections were hybridized with cRNA probes internally labeled with 35S for 16 h at 55 "C in a solution of 50% formamide, 10% dextran sulfate, 0.3 M NaCI, 1 x Denhardt's solution, 10 mM Tris-HC1, pH 8, 10 mM EDTA, and 1 x lo7 dpdrnl probe. The slides were washed with 4 x SSC for 1 h a t room temperature, and unhybridized probe was removed by digestion with a solution of 20 mg/ml ribonuclease A in 0.5 M NaCI, 0.01 M Tris-HC1, pH 8, and 0.001 M EDTA for 30 min a t 37 "C. The slides were further washed with decreasing concentrations of SSC for 1 h with a final wash in 0.1 x SSC a t 60 "C for 30 min and then dehydrated in increasing concentrations of ethanol. Slides were initially exposed to x-ray film (Hypefilm-pmax, Amersham Corp.) for 4 days to provide an indication of the intensity of the hybridization signal. They were then dipped in Kodak NTB-2 liquid autoradiography emulsion. After 10 days, the slides were developed with D-19 developer and fixed with Kodak rapid fixer.
Construction of NHE-4 Fusion Protein Antibody-A glutathione S-transferase fusion to the carboxyl amino acids 393-625 of NHE-4 was induced as described (ll), purified, and injected into New Zealand White rabbits to generate polyclonal antibodies (Cocalico Farms, Reamstown, PA).
Protein Analysis-Proteins were isolated as described previously (9). One hundred pg of membrane proteins were resolved in each lane by 7.5% SDS-PAGE. Molecular size markers run on the same gels were either Rainbow Markers(Amersham Corp.)or Bio-Rad high range standards. Proteins were then transferred to Immobilon (Millipore, Bedford, M A ) by electroelution (700-V initial setting for 1 h on Hoefer Transfor 62X, Hoefer Scientific Instruments, San Francisco, CA). The membrane was blocked at 37 "C for 1 h in TBS (0.9% NaCI, 20 mM Tris, pH 7.4) with 5% dry milk and washed three times for 5 min each in 0.1% milk, TBS. Primary NHE-4 antibody (serum) was added a t a 1:400 dilution in TBS, 1% milk, and 0.05% Tween-20 and incubated with gentle agitation a t room temperature for 2.5 h. Washes were repeated and the membranes were incubated for 30 min with donkey anti-rabbit horseradish peroxidase-conjugated secondary antibody (Jackson Immunoresearch Laboratories Inc., West Grove, PA) (1:4000 dilution in TBS, 1% milk, and 0.05% Tween-20). Washes were as above with an additional 15-min final wash. Visualization was by enhanced chemiluminescence (Amersham Corp.).
Assay (carrier-free) ,,NaCI, and the balance of the osmolarity was created by the addition of mannitol. Fluxes in all conditions were measured in the absence and presence of 500 p~ dimethylamiloride. Following the 4-min flux, cells were washed in HEPES-buffered saline and extracted in 0.5% (w/v) SDS. An aliquot was removed to count the ["Nal in a liquid scintillation counter, and protein was measured by the bicinchoninic acid procedure.

RESULTS
Localization of NHE-4 Message-In order to gain a better understanding of the potential physiological role for NHE-4, we performed in situ hybridizations in the rat kidney to determine its regional transcript expression (13). The pG7N4A probe, derived from the 5' end of NHE-4 cDNA, was constructed by subcloning the BarnHI,, to EcoRV,,, fragment into the vector pGEM7Z. This sequence of NHE-4 cDNA consists of 428 base pairs from the 5' non-coding region and 166 base pairs that encode amino acids 1-56. There is no homology between pG7N4A and any of the other cloned NHE isoforms (sequence analyses by Pustell matrix analysis). This fact was of particular importance in defining an entirely unique region of NHE-4 that would not hybridize to NHE-2, its most closely related isoform. Even at low stringency, matrix analyses could find no regions of identity with NHE-2 cDNA longer than 6 contiguous bases, a size well below the detection capabilities of the conditions used in in situ hybridization.
To verify the specificity of our pG7N4A probe, we analyzed total RNA from whole kidney and NHE-4 cDNA-transfected fibroblasts by ribonuclease protection assay. This assay is particularly sensitive to any base pair mismatches. Antisense T7generated pG7N4A cRNA was incubated with total RNA from three clonal NHE-4PS120 transfectants and from rat kidney (Fig. 1). The RNA-cRNA hybrids were incubated and washed at high stringency, and then all single-stranded RNA was digested by RNase A and T1 (same conditions of stringency as used in the in. situ hybridization analyses). A single base pair mismatch will be recognized and deleted by these RNases. The protected duplexes were visualized after separation on urea-PAGE. The 607-base probe, alone and undigested (Fig. 1, lane 1 ), includes 13 bases of vector DNA. A perfect hybrid match to NHE-4 mRNA would be 594 base pairs in size. Controls for this experiment included digested probe alone to prove that reaction conditions were sufficient to completely digest all singlestranded nucleotide, the SP6-generated sense strand probe, and total RNA from the non-transfected parental PS120 cell line. Both unhybridized probe and sense probe were completely digested, and no hybrid was detected by the antisense probe in the parental, nontransfected cells' RNA (data not shown). In those clones stably expressing NHE-4, the predicted 594-base region was protected by the NHE-4 cRNA antisense probe (Fig.  1). The same 594-base region was protected in samples of total RNA from rat kidney, indicating that the NHE-4 transcript was endogeneously expressed. I n situ hybridizations of frozen rat kidney sections with ["'SIUTP internally labeled pG7N4A antisense riboprobe revealed NHE-4 transcript primarily in the collecting ducts of the inner medulla (Fig. 2 4 ) . Tissues probed with the sense strand were clear of any hybridization signal (Fig. 2B). For both antisense and sense probes, the same conditions of high stringency hybridization and washing were used, so that all nonduplexed nucleotides were completely digested. Both sense and antisense probes were hybridized on adjacent cryosections of the same tissue. These experiments were repeated on three separate occasions on tissue freshly harvested and prepared from different animals. Fig. 2A shows distinct signals of hybridized antisense probe concentrated in the inner medulla and continuing into the outer medulla. The specificity of this probe for well delineated tubules in the inner medulla is shown in Fig. 2C. Fig. 2 0 (inner and outer medulla a t a higher magnification) shows NHE-4 message restricted to a subset of tubules, which appeared to be continuous with the collecting ducts of the inner medulla. NHE-4 message also appeared in tubules dispersed throughout the cortex (Fig. 2, E and F ) .
These probably represent cortical collecting duct segments rather than proximal tubules, the major component of the cortex, although we cannot rule out another specialized tubule expressing NHE-4 mRNA.
Expression of NHE-4 in NaIH Exchange-deficient Fibroblasts-To examine the physiological role of this isoform, we transfected a mutant NHE-deficient fibroblast cell line (Chinese hamster lung CC139 derivative, called PS120) (10) with pCN4 (full-length rat NHE-4 cDNA expressed constitutively by the cytomegalovirus promoter). This strain was mutagenized to destroy all NHE activity and has been used by many investigators to study NHE isoforms because it is one of two readily available cell lines with no endogenous NHE activity. G418resistant cells were clonally selected for genomic incorporation of the CN4 cDNA construct, and, after determining the other clonal transformants reacted similarly, one of these (PSCN4-4) was used for the more extensive studies presented here.
To confirm successful transfection of functional NHE-4 cDNA, we used both Northern and Western blot analyses of the fibroblasts. By Northern analysis (Fig. 3A), NHE-4 transcript was found only in the transfectants and not in the parent PS120 mutant fibroblasts. This indicated that the transfected NHE-4 cDNA construct was being stably transcribed into fulllength mRNA.
However, these NHE-4 transfectants expressed no amilorideinhibitable Na-H exchange activity after acid loading. The next step was to determine if the lack of NHE activity was due to a block of efficient translation or instability of the NHE-4 protein. Membrane proteins were purified from the parent cell line, PS120; PS127, a cell line expressing NHE-1; and the NHE-4 transfectants previously identified by Northern blot analysis. Immunoblot analysis (Fig. 3B) determined that only the NHE-4-transfected cell lines synthesized a 100-kDa protein that cross-reacted to the NHE-4 fusion protein antibody. Neither the parent strain (not shown) nor the PS127 transfectants expressed this protein. The apparent size by SDS-PAGE (100 kDa) is greater than the 81 kDa predicted by sequence alone, possibly due to glycosylation. NHE-4 exhibits a mobility differential similar to NHE-1, whose sequence predicted size is 91.6 kDa, but by SDS-PAGE appears to be 110 kDa. However, as shown in Fig. 3B, there is no cross-reactivity between NHE-1 and the NHE-4 antibody. This antibody also showed no crossreactivity with PS120 transfectants expressing rat NHE-1 or NHE-3 proteins (data not shown).
Effect of Hypertonicity on NHE-4 Actiuity-Next, linear influx rates of ["Nal into subconfluent PS127 (NHE-l-transfected) and PSCN4-4 (NHE-4-transfected) fibroblasts were measured over 4 min, a period where influx was determined to be linear (data not shown). Because [22Na] influxes under basal conditions were barely measurable in all cell groups, all studies were performed under acid-loaded conditions, established by 60 min of incubation with 50 mM NH,Cl, followed by a rapid washout, as described previously (12). ["Na] influxes were determined in media containing 20 mM sodium, in the presence or absence of the amiloride analog, dimethyl amiloride (DMA) (500 p~, gift from T. Kleyman, University of Pennsylvania). As shown in Fig. 4A, the NHE-deficient PS120 cells demonstrated no amiloride-inhibitable [22Na] uptake. In contrast, PS127 cells expressing NHE-1 exhibited significant ["Na] uptake, much of which was DMA-sensitive. Under the same conditions, no amiloride-sensitive [22Na] uptake could be demonstrated in the NHE-4 transfectant, PSCN4-4. These data could be consistent with PSCN4-4 cells expressing a functionally defective NHE-4 protein. However, Western analysis (Fig. 3B) indicated that a full-size NHE-4 protein was being synthesized in these cells. Given the location of NHE-4 transcript in the hyperosmolar milieu of the renal inner medullary collecting tubules, we investigated the possibility that the activation of NHE-4 might be observable only under specialized conditions. A similar Na/H exchange mechanism in barnacle muscle, which is only activated by osmolarity changes, has been described extensively (14).
We measured linear [22Nal uptake by PSCN4-4 cells acutely exposed to a wide range of extracellular osmolarities. In all  instances, Lz2Na1 uptake was measured in the presence of 20 m~ extracellular sodium and after acid loading, as described above. Osmolarity was adjusted by varying mannitol concentrations and verified by osmometry. Table I shows that there was minimal DMA-sensitive activity in NHE-4 transfected cells at osmolarities greater than 350 mosm. However, as osmolarity increased, DMA-inhibitable [22Nal uptake became demonstrable, peaking at -490 mosm, and declining thereafter (Fig. 4B). Similarly, there was an increase in NHE-1 activity in PS127 cells with increasing osmolarity, also peaking at 490 mosm. In contrast, no DMA-inhibitable activity could be measured in NHE-deficient PS120 fibroblasts at any osmolarity. Similarly, no DMA-inhibitable activity could be observed in PSCN4-4 in the absence of acidification.

DISCUSSION
This report provides the first description of the specific sites of expression for NHE-4 in the kidney and the first report of conditions under which this isoform can be activated in vitro. The location of NHE-4 transcript corresponds to a specialized region of the kidney where interstitial osmolarity is high, allowing collecting duct cells to reabsorb sodium and other solutes.
Although osmotic shrinkage has been reported to stimulate NHE activity in protein kinase C-depleted lymphocytes (15) and in human endothelial cells (16), it was not known which isoform was being affected. Based on our observations that NHE-1 is activated in acid-loaded isoosmolar and hyperosmolar conditions, we speculate these studies involved NHE-1. In contrast, we believe our observations with NHE-4 may be analogous to previous findings reporting unique activation of NHE activity in the barnacle muscle (17). NHE activity in these cells cannot be detected under basal or acid-loaded conditions, but only under conditions such as hyperosmolar fluid-induced cell shrinkage or G-protein activation. Thus, we speculate that the barnacle muscle may express an NHE-4-like exchanger. Na-H exchange in this organism may be important for cell volume regulation after hyperosmolar-induced cell shrinkage, which can occur when these cells are exposed to fluctuations in fluid tonicity encountered in sea and brackish waters. Since it appears to have little or no activity under basal or acid-loaded conditions, NHE-4, like the barnacle muscle NHE, may have a relatively minor role in cellular pH regulation.
In the PSl2O-transfected hamster fibroblasts, NHE-4 did not show any activity above 700 mosm, while NHE-1 exhibited a more gradual drop in fluxes after peaking at 490 mosm. Collecting tubules can incur external osmolarities of >lo00 mosm. If NHE-4 represents a dominant factor in regulating cell volume in the extreme conditions that occur in the kidney, we would have expected to see a much greater DMA-inhibited response in the transfected cells. Absolute flux values were also relatively low when compared to NHE-1. This discrepancy between what would be required in the in vivo milieu and what we observe in the in vitro system may be because fibroblasts do not have the regulatory or accessory proteins that are present in the renal cells and are required for maximal and sustained activation of this isoform. In contrast, NHE-1 is an ubiquitously expressed isoform, and regulatory proteins that it requires should exist in most cell types. Differences in turnover rates or post-synthetic modifications could also account for differing transport rates.
Because epithelial cells of inner medulla of the kidney are frequently subjected to significant fluctuations in luminal and interstitial osmolarity, it is possible that NHE-4 functions as a rectifying mechanism for hyperosmolar fluid-induced cell shrinkage. Its physiological role may differ substantially from NHE-1, which appears to be ubiquitous in the kidney and other tissues and has a primary role in cell pH regulation (18). NHE-1 is sensitive to and allosterically modified by increasing cellular acid load and probably plays an essential housekeeping role in most cells. NHE-1 may have an additional role in rectifying cell volume under conditions of osmolar-induced cell shrinkage. In contrast, we speculate that renal NHE-4 serves a highly specialized function, activated under conditions of hyperosmolar-induced cell shrinkage. The cells of the collecting ducts (and epithelium of the stomach and skeletal muscle, other tissues where NHE-4 is expressed) are subjected to severe mechanical and metabolic stresses, which probably require unique mechanisms to ensure efficient cell function and survival. Therefore, NHE-4 expression in areas and organisms exposed to acute and chronic changes in osmolar milieu may be of teleological significance. The mechanisms mediating cell shrinkage activation of NHE-4 remain to be explored. Because the transmembrane domains of all NHE isoforms appear structurally conserved (4), we speculate that unique differences in the predicted C-terminal, cytoplasmic domains of these proteins may determine the differences in functional and physiological roles of these isoforms.
Activation by hyperosmolarity is consistent with NHE-4 expression in collecting tubules, but this does not explain what possible role NHE-4 has in the intestine, the stomach, or other